Anthropic launched AI text watermarks
The story so far
Anthropic has introduced AI text watermarks for its latest Claude model to comply with the EU AI Act requirements, marking AI-generated content. This change covers Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6, but API users can bypass some model watermarks by adjusting sampling parameters. Anthropic states that this move is in collaboration with other major AI providers to respond to the EU market's regulations on marking AI-generated content supply.
AI-written from the reports · Updated 14 days ago
Developments
2 developments
- 1 Nov, 03:20 · 2 reportswatermark rolloutThe Decoder:Anthropic's watermark rollout leaves a loophole on two older Claude models
- 22 Oct, 18:55 · 1 reportsAI Text Watermarks in Anthropic Claude ModelsO'Reilly Radar:The Latest Anthropic Claude Models Contain AI Text Watermarks: What Does This Mean for You?
Coverage timeline
Follow the coverage to see every side of the story.
- The DecoderAnthropic's text watermark is easy to bypass on several older Claude models
Anthropic extended text watermarks to Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6, but API users can bypass the watermarks of Claude Opus 4.6 and Claude Sonnet 4.6 by adjusting the temperature, top_p, or top_k parameters.
- The DecoderAnthropic's watermark rollout leaves a loophole on two older Claude models
Anthropic extended its text watermark to Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6, but API users can bypass the watermark of two of these models by adjusting a single sampling parameter.
- O'Reilly RadarThe Latest Anthropic Claude Models Contain AI Text Watermarks: What Does This Mean for You?
Anthropic officially announced that future Claude models will include AI text watermarks to comply with the EU AI Act, including Claude Mythos 5.1 and Claude Fable 5.1.